Empath tracks changing emotions in crisis counseling talks
Empath: Tracing Multi-Level Emotion Dynamics in Crisis Counseling Dialogues
Computation and LanguageArtificial Intelligence
Summary
Understanding emotions in crisis counseling conversations is tricky because feelings change throughout the talk. The authors created Empath, a tool that looks at emotions in three ways: at each turn in the conversation, how emotions shift from one turn to the next, and overall patterns in the whole discussion. They tested Empath on chats from Black individuals dealing with grief and found patterns like ongoing negative feelings, slow moves toward hope, and different emotional roles for the person in crisis and the volunteer. This helps show how emotions unfold dynamically in crisis talks rather than staying fixed.
What this means in practice
- •For crisis support teams: Analyze ongoing emotion shifts during live chats to better tailor volunteer responses for individuals in crisis.
- •For mental health app developers: Integrate multi-level emotion tracking to enhance emotional insight features in digital mental health platforms for grief support.
Tested on one dataset.
Authors
Ziwei Gong, Yuchen Huang, Wen Liang, Nicholas Deas, Melanie Subbiah, Kathleen McKeown, Julia Hirschberg
Abstract
Emotion dynamics are critical for understanding crisis-support conversations, yet most computational work treats emotion as static utterance-level labels. We introduce EMPATH, a framework for understanding affective dynamics in mental health dialogues across three granularities: turn-level labels, transition probabilities, and global conversation archetypes. Applying EMPATH to text-based crisis conversations with self-identified Black texters discussing grief, we find persistent negative affect, gradual hope-ward transitions, distinct texter-volunteer emotional roles, and heterogeneous recovery trajectories. These results highlight the informative patterns that emerge from computationally understanding crisis support and expressions of grief as dynamic processes within conversations, as well as the overall value of emotion-dynamic analysis for analyzing and comparing affect in dialogues.